Identification and functional characterization of genes associated with lipopolysaccharide in lung adenocarcinoma

Abstract

Background and objectives

Lung cancer is one of the most common and deadliest cancers worldwide, with approximately 85% of cases classified as non-small cell lung cancer (NSCLC). Lung adenocarcinoma (LUAD) is the most prevalent subtype of NSCLC. Tumor-associated microbiota play an important role in the initiation, progression, and metastasis of tumors, but the underlying mechanisms remain unclear. Lipopolysaccharide (LPS), a major component of Gram-negative bacteria, is a key factor in inducing pathological responses. This study focuses on genes associated with lipopolysaccharide to explore the potential role of the microbiota in the pathogenesis of LUAD.

Methods

Genes associated with lipopolysaccharide were identified using the Comparative Toxicogenomics Database (CTD). Gene expression data and clinical information for LUAD were obtained from The Cancer Genome Atlas (TCGA) and Gene Expression Omnibus (GEO) databases. Univariate Cox regression analysis was performed to screen genes associated with lipopolysaccharide that were significantly associated with survival. These genes were further evaluated using Least Absolute Shrinkage and Selection Operator (LASSO) regression and multivariate Cox regression to construct a prognostic risk model. The model’s accuracy was validated using Kaplan-Meier (K-M) survival analysis, receiver operating characteristic (ROC) curves, and nomograms. We also examined the differences in immune cell infiltration, tumor mutational burden (TMB), and tumor immune dysfunction and exclusion (TIDE) between high- and low-risk groups.The expression of the key gene ITGB4 in LUAD tissues and cell lines was analyzed using qRT-PCR, WB, and IHC. The functional role of ITGB4 in LUAD cells was assessed through CCK-8, wound healing, Transwell assays, and flow cytometry. High-throughput sequencing was conducted on PC9 cells with ITGB4 knockdown to further investigate its mechanism of action. The role of ITGB4 was also validated in vivo using nude mouse xenograft models and the B6-KRAS^LSL−G12D orthotopic LUAD mouse model.

Results

Fifty-five differentially expressed genes associated with lipopolysaccharide were initially identified, among which 25 demonstrated a significant association with LUAD patient prognosis. LASSO and multivariate Cox regression analyses ultimately identified four key genes that were used to construct a prognostic model for LUAD. The results from K-M survival analysis, ROC curves, and nomograms confirmed the model’s predictive accuracy. Analysis of the tumor immune microenvironment revealed significant differences in immune cell infiltration between high- and low-risk groups. Patients in the high-risk group exhibited higher TMB and TIDE scores than those in the low-risk group. Additionally, experimental validation demonstrated that ITGB4 was highly expressed in LUAD tissues and cell lines. Knockdown of ITGB4 inhibited LUAD cell proliferation, migration, and invasion, induced G0/G1 phase arrest, and promoted apoptosis. In vivo experiments further confirmed that ITGB4 knockdown significantly suppressed tumor growth.

Conclusion

Microbiota may participate in the regulation of LUAD onset and progression through complex mechanisms. Genes associated with lipopolysaccharide provide a basis for constructing reliable prognostic models to predict outcomes in LUAD. Moreover, our study validated ITGB4 as a potential therapeutic target for LUAD at both cellular and animal levels, offering deeper insights into LUAD pathogenesis and valuable clues for therapeutic innovation.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12885-025-15387-z.

Background

Lung cancer is one of the most common types of cancer worldwide and remains the leading cause of cancer-related deaths [1]. Approximately 85% of lung cancer cases are classified as non-small cell lung cancer (NSCLC), which includes subtypes such as adenocarcinoma, squamous cell carcinoma, and large cell carcinoma. Lung adenocarcinoma (LUAD) is the predominant histological subtype of NSCLC, accounting for over 50% of all cases, and its incidence continues to rise annually [2]. Despite significant advances in treatment through surgery, chemotherapy, targeted therapy, and immunotherapy, the long-term survival rate for LUAD patients remains low, with a five-year survival rate of less than 20% [3, 4]. The occurrence and progression of lung adenocarcinoma depend on aberrant expression of key oncogenes and tumor suppressor genes, as well as dysregulation of signaling pathways [5]. Therefore, identifying novel molecular markers and effective therapeutic targets is critical for improving the prognosis and treatment outcomes of LUAD patients.

Numerous studies have extensively investigated the association between the microbiota and lung adenocarcinoma. These studies have revealed that lung adenocarcinoma tumor tissues harbor a distinct bacterial community, predominantly composed of Proteobacteria, Firmicutes, and Actinobacteria, with a microbial composition significantly different from that of adjacent normal lung tissue [6, 7]. Intratumoral microbes can activate the local immune microenvironment through metabolites (such as short-chain fatty acids) or pathogen-associated molecular patterns (e.g., LPS), thereby promoting inflammatory responses and tumor progression. A 16 S rRNA sequencing-based study demonstrated that reduced tumor microbial diversity in lung adenocarcinoma patients is significantly associated with advanced stage, EGFR mutations, and poorer prognosis [8]. Moreover, pulmonary microbial dysbiosis can upregulate PD-L1 expression, attenuate the efficacy of immune checkpoint inhibitors, and a history of antibiotic use has been linked to reduced response to immunotherapy [9]. These findings suggest that the lung adenocarcinoma tumor microbiome not only serves as a disease biomarker but may also directly participate in tumorigenesis and progression by modulating host gene expression and immune status.

With the rapid rise of research on tumor–microbiota interactions, elucidating the precise mechanisms underlying these interactions has become a major focus in contemporary oncology and holds significant potential for translational prevention and therapeutic applications [10]. Early studies have already identified tumor-specific microbial community signatures in breast, oral, prostate, and ovarian cancers [11–14]. Of particular note, microbial composition differs markedly across different stages of lung cancer, and these differences are closely correlated with lymph node metastasis, adverse prognosis, and response to immunotherapy [15]. Collectively, these observations support a key scientific hypothesis: various solid tumors, including lung adenocarcinoma, possess unique intratumoral microbial ecosystems. These ecosystems not only reflect disease status but may also function as actionable interventional targets that directly influence tumor biological behavior.

Lipopolysaccharide (LPS), a major component of the outer membrane of Gram-negative bacteria, is a key pathogenic factor. LPS can induce strong immune responses, alter the morphology, metabolism, and gene expression patterns of virtually all eukaryotic cells, promote uncontrolled expression of host cell cytokines, and trigger severe infections [16]. Numerous studies have confirmed that LPS promotes disease progression by modulating the host gene expression profile [17, 18]. The genes directly or indirectly affected by LPS, which exhibit altered expression levels, are referred to as genes associated with lipopolysaccharide.

Previous studies have shown that LPS can activate TLR4 in cancer cells, leading to the activation of NF-κB, JNK, and MAPK signaling pathways, thereby enhancing the invasive and migratory capabilities of cancer cells [17]. Moreover, LPS and LPS-induced inflammatory factors can upregulate adhesion molecules on the surface of cancer and endothelial cells, facilitating cancer cell dissemination into normal tissues. LPS from different bacterial sources exhibits marked differences in immunogenicity and biological effects, providing a novel dimension to explain the molecular heterogeneity of lung adenocarcinoma. LPS derived from Escherichia coli typically induces high levels of proinflammatory cytokines through potent activation of the TLR4/NF-κB pathway, thereby promoting tumor-associated macrophage polarization and epithelial–mesenchymal transition (EMT). This effect is particularly prone to drive an aggressive phenotype in EGFR wild-type or KRAS-mutant lung adenocarcinoma [19]. In contrast, LPS originating from Bacteroides species displays low immunogenicity due to structural modifications of lipid A (such as deacylation), which can exert antagonistic effects, suppress proinflammatory signaling, and potentially delay tumor progression [19, 20]. This microbiota source-dependent biphasic functionality of LPS suggests that the LPS composition derived from distinct bacterial communities within the same patient’s tumor may lead to differential regulation of host gene expression. Consequently, it contributes to lung adenocarcinoma heterogeneity at the levels of molecular subtype, immune microenvironment, and therapeutic response (e.g., to PD-1/PD-L1 inhibitors). These findings provide a theoretical foundation for precision stratified treatment strategies based on tumor microbiome signatures.

In this study, we investigated the potential molecular mechanisms of genes associated with lipopolysaccharide within the LUAD microbiome. Univariate and multivariate Cox regression analyses revealed that genes associated with lipopolysaccharide serve as significant prognostic indicators for LUAD patients. Based on these genes, we constructed a novel prognostic model. Additionally, we focused on a key gene, ITGB4, and further validated its role through a series of in vitro and in vivo experiments. The results showed that ITGB4 was highly expressed in LUAD cell lines and tissue samples, and its knockdown significantly inhibited the proliferation, migration, and invasion of LUAD cells, induced G0/G1 phase arrest, and promoted apoptosis. By integrating bioinformatics analysis and experimental validation, our study not only illustrates the prognostic value of genes associated with lipopolysaccharide in LUAD but also provides new insights for developing personalized treatment strategies for LUAD patients.

Methods

Data sources

We obtained mRNA expression matrix data, mutation data, and clinical data for 541 LUAD samples and 59 normal tissue samples from The Cancer Genome Atlas database (https://portal.gdc.cancer.gov/). Additionally, the LUAD dataset GSE68465 was retrieved from the Gene Expression Omnibus database (https://www.ncbi.nlm.nih.gov/geo/), which includes gene expression matrix files and clinical information for 442 LUAD samples.Through analysis of the Comparative Toxicogenomics Database (http://ctdbase.org/), we identified 154 genes associated with lipopolysaccharide in LUAD. All clinical information used in this study was obtained from the TCGA and GEO databases. Therefore, no ethical approval was required, and we strictly adhered to publication guidelines throughout the study.

Mutation frequency and copy number variation analysis

Based on data from the TCGA database, we analyzed the mutation frequency of genes associated with lipopolysaccharide using the R package “maftools.” Additionally, we assessed the amplification and deletion of copy number variations (CNVs) in genes associated with lipopolysaccharide. The chromosomal distribution of these genes was explored using the R package “RCircos.”

Construction of the prognostic risk model

After identifying genes associated with lipopolysaccharide that were significantly linked to survival via univariate Cox analysis, we further refined the selection using LASSO regression and multivariate Cox regression.Ultimately, four genes—MRPL12, LDHA, ITGB4, and NEK2—were selected to construct a prognostic risk score model. The risk score was calculated using the formula: Risk Score = Σ(expGenei × βi), where expGenei represents the relative expression level of each gene in the model, and βi is the corresponding regression coefficient.Patients were divided into high-risk and low-risk groups based on the median risk score. Kaplan-Meier survival analysis was then performed to assess overall survival (OS). The R package “pheatmap” was used to visualize patient survival status based on risk scores. Furthermore, receiver operating characteristic (ROC) analysis was conducted using the R package “timeROC,” and the area under the curve (AUC) was calculated to evaluate the predictive performance of the model.

Construction of the nomogram

In combination with clinicopathological factors, we constructed a nomogram to predict the 1-year, 3-year, and 5-year OS of patients with lung adenocarcinoma. The R packages “rms”, “regplot”, and “survival” were used to build the nomogram and its corresponding calibration curves. The closer the calibration curve is to the diagonal line, the more accurate the prediction. Additionally, the R packages “ggDCA”, “survminer”, and “survival” were used to generate decision curve analyses (DCA) to further evaluate the predictive performance of the nomogram.

Tumor-infiltrating immune cell analysis

Using the ESTIMATE algorithm, we calculated the immune scores of lung adenocarcinoma patients to reflect immune activity within the tumor microenvironment [21]. The CIBERSORT algorithm was applied to estimate the relative proportions of 22 immune cell subtypes in each tumor sample based on RNA transcript levels [22]. Additionally, we quantified and compared the relative abundances of different immune cell types between the high-risk and low-risk groups to assess and predict differences in immune cell infiltration between the two groups.

Tumor mutational burden analysis

We downloaded TMB data from the TCGA database and analyzed gene mutations in the two subgroups of lung adenocarcinoma patients using the R package “Maftools”. Differential analysis of TMB between the high-risk and low-risk groups was performed using the R packages “limma” and “ggpubr”.

Immunotherapy analysis

We obtained data on immune dysfunction and exclusion in non-small cell lung cancer using the tumor immune dysfunction and exclusion (TIDE) prediction tool (http://tide.dfci.harvard.edu/), which is used to predict responses to immunotherapy. The R packages “limma” and “ggpubr” were used to analyze TIDE scores between the high-risk and low-risk groups. TIDE scores are effective in predicting patient response to immunotherapeutic drugs; higher scores indicate a poorer response to immunotherapy [23].

Tissue sample collection and lung adenocarcinoma cell culture

All tissue specimens were obtained from the Department of Thoracic Surgery at Northern Jiangsu People’s Hospital, with approval from the hospital’s Medical Ethics Committee. Informed consent was obtained from all patients prior to sample collection. A total of 12 paired samples, including tumor tissues and matched normal tissues, were collected from lung adenocarcinoma patients who underwent tumor resection between January 2020 and December 2021. The pathological type of all samples was confirmed to be lung adenocarcinoma, and all specimens were stored at − 80 °C.The HBE, A549, H1975, H1299, and PC9 cell lines were purchased from the China National Cell Resource Center (Shanghai, China). Cells were cultured in RPMI 1640 medium (Solarbio, 31800) supplemented with 10% fetal bovine serum (FBS) (Procell, 164210–50). Cultures were incubated in a humidified incubator (Thermo Scientific, China) at 37 °C with 5% CO₂.

RNA oligo transfection procedure

Cells were seeded in 6-well plates, and transfection was initiated when cell confluency reached approximately 60%. The transfection reagent gp-transfection-mate (GenePharma) was used. All siRNAs, including the negative control siRNA (si-NC), were synthesized by GenePharma. The sequences are as follows:

• si-NC: Forward primer: 5′-UUCUCCGAACGUGUCACGUTT-3′, Reverse primer: 5′-ACGUGACACGUUCGGAGAATT-3′.

• siRNA1: Forward primer: 5′-GCGACUACACUAUUGGAUUTT-3′, Reverse primer: 5′-AAUCCAAUAGUGUAGUCGCTT-3′.

• siRNA2: Forward primer: 5′-GUGGAUGAGUUCCGGAAUATT-3′, Reverse primer: 5′-UAUUCCGGAACUCAUCCACTT-3′.

• siRNA3: Forward primer: 5′-GCUUUAAGGAAGACCACUATT-3′, Reverse primer: 5′-UAGUGGUCUUCCUUAAAGCTT-3′.

• siRNA4: Forward primer: 5′-GCUGCUUAUUGAGAACCUUTT-3′, Reverse primer: 5′-AAGGUUCUCAAUAAGCAGCTT-3′. The transfection efficiency was validated by Western blotting and RT-qPCR.

RNA extraction and quantitative real-time PCR

Total RNA was extracted from cells using TRIzol reagent (Vazyme) following the manufacturer’s instructions. RNA concentration was determined using a spectrophotometer, and the RNA samples were stored at − 80 °C for subsequent use. cDNA synthesis was performed using HiScript II qRT SuperMix (Vazyme). Quantitative real-time PCR was conducted using Hieff^® qPCR SYBR Green Master Mix (High Rox Plus) (Yeasen Biotechnology) on a real-time PCR system.The primer sequences used for qPCR were as follows:

• GAPDH: Forward primer: 5′-TCATTTCCTGGTATGACAACGA-3′, Reverse primer: 5′-GTCTTACTCCTTGGAGGCC-3′.

• ITGB4: Forward primer: 5′-GCAGCTTCCAAATCACAGAGG-3′, Reverse primer: 5′-CCAGATCATCGGACATGGAGTT-3′.

Western blot analysis

Whole-cell or tissue lysates were prepared using RIPA lysis buffer (Solarbio, R0020) supplemented with PMSF, protease inhibitors, and phosphatase inhibitor cocktail. Equal amounts of protein were separated on a 10% SDS-PAGE gel and transferred onto PVDF membranes (Immobilon-P, IPVH00010). The membranes were blocked with 5% non-fat milk and incubated overnight at 4 °C with the following primary antibodies: ITGB4 (Proteintech, 21738-1-AP), GAPDH (Proteintech, 10494-1-AP), MMP2 (Proteintech, 10373-2-AP), MMP9 (Proteintech, 10375-2-AP), p53 (Proteintech, 10442-1-AP), Caspase-3 (Cell Signaling Technology, 9664 T), and Bcl-2 (Cell Signaling Technology, 4223 T). After washing, the membranes were incubated with HRP-conjugated secondary antibody (IgG, ABclonal, AS014) at room temperature for 1 h.Protein bands were visualized using the Super ECL Detection Reagent (Yeasen Biotechnology), and band intensities were quantified by ImageJ software.

Immunohistochemical staining and scoring

Paraffin-embedded lung adenocarcinoma tissue sections were deparaffinized and rehydrated, followed by antigen retrieval in a boiling water bath containing sodium citrate buffer for 30 min. The sections were then incubated in endogenous peroxidase blocking solution in the dark for 15 min, followed by incubation in immunostaining blocking buffer for another 15 min to block nonspecific binding.Subsequently, the sections were incubated overnight at 4 °C with the primary antibody ITGB4 (Proteintech, 21738-1-AP). After returning to room temperature, the sections were incubated with an HRP-conjugated secondary antibody (IgG, Servicebio, G1215-200T) at room temperature for 1 h. DAB chromogen was used for signal development, followed by counterstaining and differentiation with a hematoxylin-eosin staining kit (Solarbio, G1120). The slides were then dehydrated and mounted.The scoring criteria for ITGB4 expression were as follows: Staining intensity score:0 (no staining),1 (light brown),2 (brown),3 (dark brown). Percentage of positive cells: 0 (0–.5%), 1 (6–25%), 2 (26–50%), 3 (51–75%), 4 (> 75%). The final immunohistochemistry (IHC) score was calculated by multiplying the staining intensity score by the percentage of positive cells, with a total score ranging from 0 to 12. A score of ≤ 6 was defined as low expression, while a score of > 6 was defined as high expression.

Cell proliferation assay

Cell proliferation was assessed using a 96-well plate. After cell counting, 1,000 cells were seeded into each well. At 24, 48, 72, and 96 h, 10 µL of CCK-8 solution (Yeasen) was added to each well, followed by incubation in the dark for 90 min. The absorbance at 450 nm was measured using a microplate reader (SkanIt RE 7.0).

Cell migration and invasion assays

Cell migration assay was performed using 8-µm pore size Transwell chambers (Corning, USA) placed in 24-well plates. The upper chamber was loaded with 200 µL of serum-free cell suspension containing 10,000 cells, while the lower chamber was filled with 500 µL of complete medium supplemented with 10% FBS. Cells were incubated for 48 h in a cell culture incubator. Following incubation, chambers were rinsed with ddH₂O to remove debris. Cells were then fixed with 4% paraformaldehyde for 10 min, stained with 0.1% crystal violet solution for 7 min, and washed three times with ddH₂O. Non-migrated cells on the upper surface (non-migratory side) of the membrane were gently removed using a cotton swab, and chambers were air-dried at room temperature. Migrated cells adhering to the lower surface (lower chamber side) were examined for quantity and morphology under an inverted microscope (OLYMPUS-CKX53) and photographed. Migrated cell counts were quantified using ImageJ software, and experimental results were expressed as mean values.

Wound healing experiment

Transfected cells were seeded in 6-well plates and cultured until confluent monolayers formed. A straight wound was created in each monolayer using a sterile pipette tip. After washing with PBS to remove detached cells, cells were further incubated in serum-free medium. Wound closure was monitored at 0, 24, and 48 h using an inverted microscope (OLYMPUS-CKX53), with images captured at each time point. Quantification of wound healing was performed by measuring wound width using ImageJ software.

Flow cytometry analysis

Cell Cycle Detection: Cells were washed twice with PBS, digested with trypsin, and collected. The cells were fixed overnight with 70% ethanol at 4 °C. Fixed cells were centrifuged and resuspended, then 0.5 ml of Propidium Iodide staining solution (Beyotime) was added. The mixture was incubated at 37 °C in the dark for 30 min.Cell Apoptosis Detection: After washing twice with PBS and trypsin digestion, cells were resuspended in pre-cooled binding buffer. Annexin V-FITC reagent (Beyotime) was added, and the cells were incubated at 20 °C in the dark for 20 min. All samples were detected by flow cytometry (BD Biosciences), and data were analyzed using FlowJo software.

Reference-based transcriptome sequencing analysis

Total RNA extraction, high-throughput sequencing, and raw data processing were commissioned to Shanghai OE Biotech Co., Ltd. Cells were divided into siNC and siRNA3 groups (3 samples per group) for RNA oligo transfection. After 48 h of culture, cells were collected, and total RNA was extracted using TRIzol. Subsequently, the VAHTS Universal V5 RNA-seq Library Prep Kit was used to construct libraries, which were sequenced on an Illumina NovaSeq 6000 platform to generate 150 bp paired-end reads. Raw reads were quality-controlled by fastp to obtain clean reads [24], which were then aligned to the reference genome using HISAT2 for FPKM calculation [25, 26]. Gene read counts were quantified by HTSeq-count [27]. Differentially expressed genes were screened using DESeq2 (q-value < 0.05, |log₂FC| >1) [28], followed by Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment analyses based on the hypergeometric distribution algorithm [29, 30]. Volcano plots, bar charts, and bubble plots were generated using R software (version 3.2.0).

Xenograft model

PC9 cells were seeded into 10 cm culture dishes and transfected with siNC or siRNA3, forming two experimental groups. A total of 10 BALB/c nude mice (purchased from Nanjing GemPharmatech Co., Ltd.) were randomly divided into two groups. Subsequently, 9 × 10⁵ treated cells were subcutaneously injected into the axillary region of each mouse. Tumor volumes were measured every 7 days using the formula: Volume = (Length × Width²) × 0.5. After 3 weeks, the mice were euthanized via cervical dislocation, and the subcutaneous tumors were promptly harvested for volume and weight measurement.

KRAS orthotopic lung cancer mouse model

KRAS conditional mutant C57BL/6J mice (B6-KRAS^LSL−G12D), aged 6–8 weeks, were purchased from Nanjing GemPharmatech Co., Ltd. During the experiment, the mice were anesthetized with isoflurane inhalation and then infected via intratracheal instillation of 2.5 × 10⁷ PFU adenovirus (Ad-Cre, OBIO). The mice were sacrificed 6 months post-infection. All animal experiments were conducted in strict accordance with animal welfare guidelines and were approved by the Experimental Animal Ethics Committee of Yangzhou University.

Statistical methods

Statistical analyses were performed using GraphPad Prism 8.0 and R software (version 4.2.2). Differences between two groups were assessed using Student’s t-test, while comparisons among multiple groups were conducted using one-way ANOVA. A P-value less than 0.05 was considered statistically significant.

Results

Identification and characterization of lipopolysaccharide-related genes in lung adenocarcinoma

To identify differentially expressed genes associated with lipopolysaccharide in LUAD, we analyzed 541 LUAD samples and 59 normal tissue samples from the TCGA database.Differential expression analysis was conducted using the limma package with thresholds set at |log2FC| ≥ 1 and p < 0.05. This analysis revealed 5,145 differentially expressed genes between LUAD and normal tissues. We then intersected these genes with 154 previously identified genes associated with lipopolysaccharide, yielding 55 overlapping genes (Fig. 1A).Further univariate Cox regression analysis identified 25 prognosis-related genes (Fig. 1B). Box plots illustrate the expression levels of these genes associated with lipopolysaccharide in lung adenocarcinoma and normal tissues (Fig. 1C).

Fig. 1.

Identification of genes associated with lipopolysaccharide that are linked to prognosis in lung adenocarcinoma. A Venn diagram showing the overlap between differentially expressed genes and genes associated with lipopolysaccharide. B Univariate Cox regression analysis of genes associated with lipopolysaccharide. C Expression levels of the prognosis-associated genes associated with lipopolysaccharide in tumor versus normal tissues

To further investigate the genetic alteration characteristics of genes associated with lipopolysaccharide in lung adenocarcinoma, we analyzed somatic mutation data from 390 LUAD samples in the TCGA database. We found that 70 samples (17.95%) harbored statistically significant mutations in genes associated with lipopolysaccharide, with the top three mutated genes being MET (3%), MMP1 (2%), and PRAME (2%) (Supplementary Fig. 1A). Copy number variation analysis of genes associated with lipopolysaccharide revealed CNV events in a subset of samples, including copy number amplification of MRPL12, ITGB4, and KPNA2, as well as copy number deletion of CCNB1 and GAPDH (Supplementary Fig. 1B). The chromosomal locations of the genes associated with lipopolysaccharide are shown in Supplementary Fig. 1C.

Construction and evaluation of the risk prediction model

To assess the prognostic potential of genes associated with lipopolysaccharide in lung adenocarcinoma, we developed a risk prediction model. First, 25 genes significantly associated with prognosis were identified through univariate Cox regression analysis. These genes were further refined using LASSO regression analysis, resulting in seven candidate genes (Supplementary Fig. 2). Ultimately, multivariate Cox regression analysis identified four key genes associated with lipopolysaccharide—MRPL12, LDHA, ITGB4, and NEK2—as hub genes, which were subsequently used to construct the prognostic model. The risk score was calculated using the following formula: Risk score = (MRPL12 × 0.179250852208785) + (LDHA × 0.513602401787825) + (ITGB4 × 0.147373557644523) + (NEK2 × 0.154043440358349). We used the TCGA dataset as the training set and the GSE68465 dataset as the validation set. Kaplan–Meier survival analysis was performed for different risk groups (Fig. 2A), and the results consistently showed that patients in the high-risk group had significantly worse prognosis in both the training and validation cohorts. As the risk score increased, patient mortality also rose accordingly (Fig. 2B). Furthermore, we assessed the predictive performance of the risk model using ROC curve analysis. The AUC values for the training set at 1, 3, and 5 years were 0.689, 0.708, and 0.652, respectively, while the corresponding AUC values in the validation set were 0.714, 0.672, and 0.623 (Fig. 2C), confirming the predictive accuracy of the model.

Fig. 2.

Evaluation of the risk prognostic model. A Kaplan–Meier survival curves for different risk groups. B Distribution of risk scores and survival status in the training and validation sets. C ROC curves for predicting 1-, 3-, and 5-year overall survival

Development and validation of the nomogram

Multivariate Cox regression analysis demonstrated that the risk score was an independent prognostic factor for overall survival in lung adenocarcinoma patients in both the training set and validation set (HR > 1, both P < 0.001; Supplementary Fig. 3A–B). A nomogram for lung adenocarcinoma patients was constructed based on sex, age, clinical stage, and risk score (Supplementary Fig. 3 C). The calibration curves showed that the predicted 1-, 3-, and 5-year overall survival rates were in good agreement with the actual observed outcomes (Supplementary Fig. 3D). Decision curve analysis for 1-, 3-, and 5-year survival demonstrated that the nomogram provided more accurate survival predictions than models relying solely on individual clinical parameters such as age, sex, and stage (Supplementary Fig. 3E).

Assessment of the tumor immune microenvironment and immune cell correlation based on the risk model

To explore differences in immune cell infiltration between high- and low-risk groups, we applied the ESTIMATE algorithm and found that the low-risk group exhibited significantly higher ESTIMATE scores and immune scores compared to the high-risk group (Supplementary Fig. 4A). Further analysis using the CIBERSORT algorithm revealed significant differences between the two groups in the infiltration levels of memory B cells, resting CD4 memory T cells, activated CD4 memory T cells, M0 macrophages, M1 macrophages, and resting mast cells (P < 0.001) (Supplementary Fig. 4B). We also analyzed the correlation between the four key genes, the risk score, and various immune cell types, and further investigated the relationship between the risk score and immune cell infiltration (Supplementary Fig. 4C). The results showed that the risk score was significantly associated with the infiltration levels of memory B cells, M0 macrophages, M1 macrophages, activated mast cells, resting mast cells, neutrophils, resting CD4 memory T cells, and activated CD4 memory T cells (P < 0.001) (Supplementary Fig. 4D).

Tumor mutational burden analysis and immunotherapy response evaluation based on the risk model

We used the Maftools algorithm to analyze somatic mutations in patients from the high- and low-risk groups. The results revealed a broader spectrum of somatic mutations in the high-risk group. For example, the mutation frequency of TP53 was 51% in the high-risk group compared to 35% in the low-risk group; TTN mutations occurred in 54% vs. 32%, and MUC16 mutations in 43% vs. 35%, respectively (Supplementary Fig. 5 A). The TMB was significantly higher in the high-risk group than in the low-risk group, with statistical significance (p < 0.001) (Supplementary Fig. 5B). To further evaluate the potential response to immunotherapy, we applied the TIDE algorithm. A higher TIDE score indicates a greater likelihood of immune evasion, suggesting a lower potential benefit from immunotherapy. In our study, TIDE scores differed significantly between the two risk groups, with the low-risk group showing lower TIDE scores, indicating that these patients may benefit more from immunotherapy (Supplementary Fig. 5C).

ITGB4 expression is significantly upregulated in lung adenocarcinoma cells and tissues

ITGB4 was selected as the candidate gene for further validation due to its most pronounced differential expression and prognostic significance in lung adenocarcinoma. Analysis using the GEPIA2 database revealed that ITGB4 expression was significantly higher in 483 lung adenocarcinoma tissues than in 347 normal lung tissues (Fig. 3A). Kaplan–Meier survival analysis further demonstrated that patients with high ITGB4 expression had significantly worse prognosis than those with low expression (Fig. 3B).To validate these findings, we examined ITGB4 expression in lung cancer cell lines (A549, H1299, PC9, H1975) and the normal human bronchial epithelial cell line HBE (a widely used non-malignant control derived from primary human bronchial epithelial cells immortalized with SV40 large T antigen) by qRT-PCR and Western blot. The results showed that ITGB4 expression was markedly elevated in lung adenocarcinoma cell lines compared to HBE, whereas expression in H1299 was relatively lower, consistent with its classification as large-cell carcinoma rather than adenocarcinoma (Fig. 3C, D). Inclusion of H1299 was intended to provide broader representation across non-small cell lung cancer subtypes for comparative analysis. Additionally, ITGB4 protein levels were analyzed in 12 paired LUAD tissues and adjacent normal tissues, showing a significant upregulation of ITGB4 in tumor tissues (Fig. 3E). Immunohistochemistry further confirmed the high expression of ITGB4 in LUAD tissues, with predominant localization on the cell membrane (Fig. 3F).

Fig. 3.

Expression of ITGB4 in Lung Adenocarcinoma Cells and Tissues. A Comparison of ITGB4 expression between 483 LUAD samples and 347 normal samples. B K–M survival curves for high and low ITGB4 expression groups. C–D ITGB4 expression levels in normal lung epithelial cell lines and LUAD cell lines analyzed by qRT-PCR and WB. E WB analysis of ITGB4 expression in 12 paired LUAD tissues. F IHC detection of ITGB4 expression in LUAD tissues. *p < 0.05, **p < 0.01

Silencing ITGB4 suppresses tumor cell proliferation, migration, and invasion

Both A549 and PC9 cells exhibited high expression levels of ITGB4. Therefore, siRNA transfection was performed in these two cell lines to knock down ITGB4, and the knockdown efficiency was verified by qRT-PCR and WB analysis (Fig. 4A). siRNA2 and siRNA3 were selected for subsequent functional assays. Results from the CCK-8 assay showed that silencing ITGB4 significantly inhibited the proliferation of A549 and PC9 cells (Fig. 4B). Wound healing and Transwell assays further confirmed that ITGB4 knockdown markedly reduced the migratory and invasive capacities of the cells (Fig. 4C and E). In addition, we observed that the expression levels of MMP-2 and MMP-9—were significantly decreased following ITGB4 knockdown (Fig. 4D).

Fig. 4.

Knockdown of ITGB4 Inhibits Proliferation, Migration, and Invasion of Lung Adenocarcinoma Cells A.Efficiency of ITGB4 knockdown in A549 and PC9 cells validated by qRT-PCR and WB. B.Cell proliferation assessed using the CCK-8 assay. C.Cell migration evaluated by wound healing assay. D.Expression levels of migration-related proteins (MMP2, MMP9) detected by WB. E. Transwell assay used to assess cell migration and invasion abilities. ***p < 0.001, ****p < 0.0001

ITGB4 knockdown induces G1 phase arrest and promotes apoptosis

We used flow cytometry to investigate whether ITGB4 knockdown induces cell cycle arrest and increases apoptosis in lung adenocarcinoma cells. In PC9 cells, compared with the siNC group, cells transfected with siRNA3 targeting ITGB4 exhibited a significant increase in the proportion of cells in the G0/G1 phase and a marked decrease in the S phase population (Fig. 5A). Further analysis revealed that ITGB4 knockdown significantly increased the rate of apoptosis (Fig. 5B). In addition, we examined the expression of apoptosis-related proteins and found that silencing ITGB4 led to a marked upregulation of Caspase-3 and P53, along with a significant downregulation of Bcl-2 (Fig. 5C), further confirming that ITGB4 knockdown promotes apoptosis.

Fig. 5.

ITGB4 Knockdown Induces G1 Phase Arrest and Promotes Apoptosis. A Effects of ITGB4 knockdown on the cell cycle of PC9 cells. B Effects of ITGB4 knockdown on apoptosis in PC9 cells. C Changes in the expression of apoptosis-related proteins (Caspase-3, P53, Bcl-2) in PC9 cells. ***p < 0.001

Functional and pathway analysis of ITGB4 regulation

To further explore the potential mechanisms by which ITGB4 functions in lung adenocarcinoma cells, we performed high-throughput sequencing and differentially expressed gene (DEG) analysis on PC9 cells following ITGB4 knockdown. A total of 622 DEGs were identified, including 159 upregulated and 463 downregulated genes (Fig. 6A). These DEGs were subsequently subjected to GO and KEGG enrichment analyses. GO enrichment analysis indicated that ITGB4 is associated with biological processes such as neutrophil chemotaxis and inflammatory response, cellular components such as cornified envelope and extracellular region, and molecular functions including CCR chemokine receptor binding, chemokine activity, and pattern recognition receptor activity (Fig. 6B). KEGG pathway analysis revealed that the upregulated DEGs were primarily enriched in metabolic pathways such as steroid biosynthesis and unsaturated fatty acid biosynthesis, suggesting a potential role of ITGB4 in the regulation of cellular metabolism (Fig. 6C). Conversely, the downregulated DEGs were mainly involved in signaling pathways related to interactions between viral proteins and host cytokines and receptors, as well as cytokine–cytokine receptor interactions (Fig. 6D). This study, through the integration of high-throughput sequencing and bioinformatics analysis, reveals the functions of ITGB4 and its potential molecular mechanisms in lung adenocarcinoma.

Fig. 6.

Functional and Pathway Analysis Regulated by ITGB4. A Volcano plot of differentially expressed genes. B GO enrichment analysis. C–D KEGG enrichment analysis

Pro-tumorigenic role of ITGB4 in tumor development in vivo

To investigate the in vivo function of ITGB4, we established a xenograft model using nude mice. PC9 cells transfected with either siNC or siRNA3 were injected into the mice, and tumor growth was continuously monitored. The results showed that ITGB4 knockdown significantly suppressed tumor growth (Fig. 7A), with notable reductions in both tumor volume and weight (Fig. 7B and C). Additionally, an orthotopic lung cancer model was established in mice, and both tumor tissues and adjacent normal tissues were collected for analysis. IHC staining revealed that ITGB4 expression was significantly elevated in lung cancer tissues compared to normal lung tissues (Fig. 7D), further supporting the conclusion that ITGB4 plays a pro-tumorigenic role in the progression of lung adenocarcinoma.

Fig. 7.

ITGB4 Promotes Tumor Growth In Vivo. A Xenograft model in nude mice. B Tumor volume measurement. C Tumor weight evaluation. D ITGB4 expression levels detected in the orthotopic lung cancer model by IHC. **p < 0.01, ***p < 0.001, ****p < 0.0001

Discussion

Due to current technical limitations, the diversity of the tumor microbiome and its impact on tumor progression remain insufficiently understood. Studies have shown that microorganisms participate in multiple stages of tumor development, including initiation, progression, metastasis, and immune response, with a particularly notable role in LUAD. Therefore, applying bioinformatics approaches to investigate the potential mechanisms of the microbiome in LUAD has become an important research direction.

In this study, genes associated with lipopolysaccharide, as identified from the CTD database, were analyzed through bioinformatics approaches to systematically investigate their potential role in LUAD.A prognostic risk model and nomogram were constructed, and four key genes—MRPL12, LDHA, ITGB4, and NEK2—were identified. LDHA, a key enzyme in glycolysis, catalyzes the conversion of pyruvate to lactate [31, 32]. Its high expression has been observed in various cancers and is closely associated with the Warburg effect, whereby cancer cells preferentially undergo glycolysis even in the presence of oxygen [33]. Under LPS-induced inflammatory conditions, LDHA may promote tumor cell survival and proliferation by enhancing glycolysis. ITGB4 plays a crucial role in cell adhesion to the extracellular matrix and is essential for cell migration, proliferation, and survival. It is highly expressed in several malignancies and has been strongly associated with poor prognosis in cancers such as breast, cervical, head and neck, and pancreatic cancers [34–40]. Its oncogenic effects are thought to be mediated through the α6β4 integrin complex, which promotes proliferative signaling, apoptosis resistance, invasion, metastasis, and angiogenesis [41]. NEK2, a serine/threonine kinase, is overexpressed in various human cancers and is involved in tumor development and immune evasion [42–44]. It has been shown to stabilize PD-L1 expression, thereby suppressing antitumor immunity [42, 45]. NEK2 inhibition not only enhances the efficacy of PD-L1 blockade but also induces cell cycle arrest, suggesting that targeting NEK2 may synergize with chemotherapy, radiotherapy, targeted therapy, and immunotherapy as a promising cancer treatment strategy [42]. MRPL12 is significantly upregulated in hepatocellular carcinoma and promotes the malignant phenotype by modulating mitochondrial metabolism, highlighting its role in metabolic reprogramming [46].

Mounting evidence indicates that LPS derived from Gram-negative bacteria represents one of the most critical exogenous pro-carcinogenic signals in the lung adenocarcinoma microenvironment. LPS activates the NF-κB, MAPK, and PI3K/AKT pathways via TLR4, thereby not only reshaping an immunosuppressive microenvironment but also directly upregulating ITGB4 expression and enhancing its function through Src-mediated tyrosine phosphorylation [47–52]. Consequently, the high expression and hyperactivation of ITGB4 are, to a large extent, an active outcome of regulation by the microbiota-LPS axis rather than a purely autonomous event of cancer cells. This mechanism precisely underlies the core reason why the present prognostic model can effectively capture microbiota-associated malignant phenotypes.

Immunotherapy is considered one of the most significant breakthroughs in oncology. Despite its progress, the benefits of immunotherapy remain limited to specific patient populations and tumor types [53]. This study further investigated immune and genomic features associated with different risk stratifications and examined the relationship between genes associated with lipopolysaccharide and immune responses. The results showed that the expression of genes associated with lipopolysaccharide was significantly correlated with the infiltration of various immune cells, including memory B cells, resting and activated CD4 memory T cells, M0 and M1 macrophages, and resting mast cells. In addition, TMB, defined as the number of somatic mutations per million base pairs, is commonly used as a predictive biomarker for immune checkpoint blockade in lung cancer [54]. Our analysis revealed that patients in the high-risk group exhibited a significantly higher TMB compared to those in the low-risk group, along with a higher mutation frequency in key genes such as TP53, TTN, and MUC16. TP53 is the most commonly mutated gene in LUAD and serves as a potential biomarker for immune checkpoint inhibitor therapy [55]. TTN mutations, also found across multiple solid tumors, and MUC16 overexpression, commonly seen in various cancers, have been associated with early diagnosis and prognosis of lung cancer [56–58]. TIDE, a computational framework for predicting immunotherapy response [23], indicated that patients in the high-risk group had higher TIDE scores, suggesting a greater likelihood of immune evasion.

We further evaluated the role of ITGB4 in LUAD. Patients with high ITGB4 expression had significantly shorter overall survival compared to those with low expression. Functional assays, including CCK-8, wound healing, Transwell migration/invasion, and flow cytometry, demonstrated that ITGB4 knockdown significantly inhibited the proliferation, migration, and invasion of A549 and PC9 cells while promoting apoptosis.One of the key drivers of cancer progression is the dysregulation of cell migration and apoptosis, processes governed by multiple oncogenic pathways [58–61]. Cancer cell migration and invasion into surrounding tissues and the vasculature are essential steps in metastasis—the primary cause of cancer-related death [62]. Migration involves complex changes, including cytoskeletal remodeling, reduced adhesion, and loss of polarity [63], often marked by increased expression of mesenchymal markers and alterations in matrix metalloproteinases (e.g., MMP2, MMP7, MMP9) [64, 65]. Apoptosis is mediated via intrinsic and extrinsic pathways, with proteins such as Bcl-2 and Caspase-3 being key regulators [66]. The tumor suppressor p53 can also promote apoptosis through transcription-independent mechanisms [67]. In our study, ITGB4 downregulation led to decreased MMP2 and MMP9 expression, increased Caspase-3 and p53 levels, and reduced Bcl-2 expression.

GO enrichment analysis showed that ITGB4 is involved in processes related to cell migration and inflammatory responses. Cancer cells acquire migratory capabilities by remodeling the cytoskeleton and regulating adhesion molecules such as integrins. Chronic inflammation is a recognized promoter of cancer, contributing to DNA damage, enhanced proliferation, and apoptosis inhibition. KEGG enrichment analysis further revealed that ITGB4 participates in metabolism-related pathways. Given that cancer is a metabolic disease characterized by increased energy demands, tumor cells often reprogram metabolism to support rapid growth and survival [68].In vivo, we found that ITGB4 knockdown significantly suppressed tumor growth in a xenograft mouse model. In a B6-KRAS^LSL−G12D mouse model of LUAD, immunohistochemistry showed significantly elevated ITGB4 expression in tumor tissues compared to adjacent normal tissues, indicating a potential oncogenic role of ITGB4 in LUAD development.

Despite providing valuable insights into the molecular mechanisms of LUAD, particularly the potential role of ITGB4, this study has certain limitations. Firstly, the present study primarily relies on retrospective data from public databases, which may introduce bias due to inconsistencies in data curation, limited sample diversity, and incomplete representation of dynamic biological processes, thereby constraining the generalizability of the findings. Additionally, the study relies heavily on genes associated with lipopolysaccharide curated in the CTD database, which may not comprehensively capture the full range of microbiota–host interactions. Secondly, although the prognostic model was externally validated using the GSE68465 dataset and the biological functions of ITGB4 were preliminarily investigated through in vitro and in vivo experiments, the remaining three genes associated with lipopolysaccharide (MRPL12, LDHA, and NEK2) have not undergone similar experimental validation, leaving their direct roles in the pathogenesis of lung adenocarcinoma uncertain. Furthermore, the xenograft model employed transient siRNA transfection prior to implantation, resulting in a limited knockdown duration (< 10 days). Consequently, the observed long-term tumor suppression may partly stem from persistent biological effects induced by early silencing. Future studies should utilize shRNA-mediated stable knockdown for more definitive validation. Moreover, the precise regulatory mechanisms of ITGB4 within the complex tumor microenvironment of lung adenocarcinoma remain incompletely elucidated. Given the inherent limitations of cell lines and animal models—including discrepancies with human physiology and intratumoral heterogeneity—the clinical translation of these findings should be interpreted with caution. Therefore, future studies should include systematic mechanistic research and large-scale prospective clinical validation to clarify the functional roles and therapeutic potential of ITGB4 and related genes in LUAD.

In summary, we established a robust prognostic model for lung adenocarcinoma based on four genes associated with lipopolysaccharide. This model demonstrated superior predictive accuracy for overall survival and showed strong correlations with immune cell infiltration, tumor mutational burden, and response to immunotherapy. Notably, ITGB4 was significantly upregulated in lung adenocarcinoma, promoting tumor cell proliferation, migration, and invasion, thereby emerging as a potential oncogene linked to adverse prognosis and a promising prognostic biomarker and therapeutic target. This work provides novel theoretical support for immunotherapy and personalized treatment in lung adenocarcinoma, with considerable clinical translational potential, including the development of ITGB4-specific inhibitors (small molecules or antibodies) to abrogate its oncogenic activity, microbiota-modulating interventions (e.g., probiotics or antibiotics) to attenuate LPS-driven ITGB4 upregulation, and integration of the risk model into clinical decision-making to stratify patients who may benefit from immune checkpoint inhibitors or combination regimens. Future studies are warranted to further elucidate the underlying mechanisms and validate these strategies in prospective clinical trials.

Abbreviations

NSCLC

Non-small cell lung cancer

LUAD

Lung adenocarcinoma

LPS

Lipopolysaccharide

CTD

Comparative toxicogenomics database

TCGA

The Cancer Genome Atlas

GEO

Gene Expression Omnibus

LASSO

Least Absolute Shrinkage and Selection Operator

K- M

Kaplan-Meier

ROC

Receiver operating characteristic

TMB

Tumor mutational burden

TIDE

Tumor immune dysfunction and exclusion

CNV

Copy number variation

OS

Overall survival

AUC

The area under the curve

DCA

Decision curve analyse

FBS

Fetal bovine serum

IHC

Immunohistochemistry

DEG

Differentially expressed gene

GO

Gene Ontology

KEGG

Kyoto Encyclopedia of Genes and Genomes

Authors’ contributions

SDZ: Conceptualization, Supervision, Formal analysis, Writing-original draft, Writing-review & editing. MJG, SYD: Formal analysis, Writing–review & editing. MMW, HBX: contributed to discussions and suggestions. YSS, XLW: Reviewed and approved the final version of the manuscript.

Funding

This work was supported by Jiangsu Provincial Health Commission Elderly Health Research Project [No. LKZ2022019].

Data availability

The datasets used in this study are available from the corresponding author upon reasonable request.

Declarations

Ethics approval and consent to participate

This study was approved by the Ethics Committee of Northern Jiangsu People’s Hospital (Approval No. 2021ky012-1), and written informed consent was obtained from all participants prior to enrollment. The animal experimental protocols were approved by the Experimental Animal Ethics Committee of Yangzhou University (Ethics No. yzu-lcyxy-s036). All experimental procedures were conducted in strict accordance with relevant guidelines and regulations, and the study was carried out in compliance with the ARRIVE guidelines.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.